Your AI QA agent after every deploy
In plain terms: an agent tests your live app, finds what's broken, and hands you the fix — ready to paste into your editor.
Paste a URL — agents explore the app, click every button, test forms and real user flows, then write a fix prompt ready for Claude Code, Cursor or any editor. Built for vibe-coded apps; works for any site.
No signup · instant preview · full report + fix prompt are free with an account
- No credit card
- Non-destructive checks
- Credentials encrypted, never stored
- Fix prompt ready for your editor
How it works
From URL to fix prompt in one pass
No setup, no scripts. Agent Smith treats every deploy like a release candidate.
Paste your URL
Drop the link to your deployed app. Add optional login, IP/API-key whitelist notes and special instructions.
Agents go to work
A swarm of QA agents explores pages, clicks buttons, tests forms and records every error, slow load and weak spot.
Get an actionable report
Scores, prioritised findings, screenshots, UX suggestions — and a fix prompt ready to paste into Claude Code or Cursor.
The agent swarm
Seven specialised agents, one report
Each agent owns a dimension of quality. Together they cover what a human QA pass would — automatically.
ExplorerAgent
Discovers pages, links, buttons and forms — maps your whole surface.
FunctionalQAAgent
Tests buttons & forms, catches console + network errors and bad routes.
UXAgent
Scores clarity, friction, onboarding, empty states and visual hierarchy.
UIAgent
Checks responsive, contrast, alt text and design consistency.
SecurityLightAgent
Passive review of headers, cookies, CSP & HTTPS. No aggressive scanning.
PerformanceAgent
Measures load times, flags slow pages and failing network requests.
PromptFixAgent
Turns every finding into one ready-to-paste fix prompt for your AI IDE.
Sample report
The report you actually act on
Scores per dimension, prioritised findings, and a fix prompt engineered for your AI IDE.
Target
my-saas.app
74
/ 100
Fix prompt for Claude Code / Cursor
You are a senior engineer working on the project behind my-saas.app. An automated QA audit (Agent Smith) found the issues below. Fix them WITHOUT breaking existing functionality, in priority order. ### 1. [HIGH · security] Missing 5 recommended security headers - Fix: add CSP, HSTS, X-Frame-Options, X-Content-Type-Options, Referrer-Policy. ### 2. [HIGH · functional] JavaScript console errors on /dashboard - Fix: guard undefined access in the projects list; wrap the fetch in try/catch. ### 3. [MEDIUM · performance] /dashboard loads in 4.1s - Fix: code-split the chart bundle and lazy-load below-the-fold widgets.
Paste straight into your AI IDE and let it fix the issues in one pass.
Agent marketplace
Hire specialised agents on demand
Beyond the core, plug in named agents tuned for UX, security, mobile or conversion.
Agent Smith
Full-spectrum autonomous QA
Crawls the whole surface, stress-tests flows, and compiles a single actionable report. The baseline every audit runs on.
Agent Mouni
UX / QE with human review
Pairs automated exploration with heuristic UX scoring — onboarding clarity, friction points, empty states and microcopy.
Agent Karim
Passive security review
Inspects headers, cookies, CSP, auth surfaces and error leakage. Strictly non-destructive — no brute force, no exploitation.
Agent Lina
Responsive & mobile QA
Replays flows across phone, tablet and desktop viewports. Catches tap targets, overflow, viewport bugs and layout shift.
Agent Néo
Stratégie produit / growth
Lit le produit comme un partner growth : proposition de valeur, activation (time-to-wow), rétention, monétisation, acquisition/viralité, confiance et mesure. Rend un plan priorisé par levier, pas une liste de bugs.
Agent Tank
Analytics / mesure (growth-ops)
Détecte si tu mesures vraiment ton trafic : d'où viennent les users, quels canaux, quel funnel, quelles conversions. Dit quoi instrumenter et propose de monter ton dashboard de visibilité directement sur Agent Smith.
Agent Link
SEO / GEO-AEO (référencement)
Audit SEO Google (mots-clés, on-page, technique, mots-clés des concurrents) ET GEO/AEO : être cité par les LLM (ChatGPT, Claude, Perplexity) via llms.txt, données structurées et contenu citable. Analyse les vrais signaux de la page.
Agent Morpheus
Direction / CEO (vision exécutive)
Raisonne comme le fondateur : LA priorité du trimestre, ce qu'il faut construire/tuer/reporter, le marché, l'avantage défendable (moat), les unit economics, le risque #1 et le pari à fort upside. Tranché.
Agent Trinity
Sales / CRO (taux de conversion)
Regarde le site sous l'angle VENTE : clarté du hero, CTA, preuve sociale, tunnel, friction, pricing, urgence. Dit concrètement « mets ça plutôt que ça » et propose des tests A/B pour convertir plus.
Agent Oracle
Design / UI (direction artistique)
Juge le craft visuel : hiérarchie, espacement, typographie, palette/contraste, cohérence, profondeur, états et motion. Propose des améliorations graphiques précises (valeurs, échelles, do/don't).
Agent Theo
Funnel / conversion QA
Walks signup, checkout and activation funnels end-to-end, flagging drop-off traps, dead CTAs and trust gaps.
Pricing
Start free. Scale when it ships value.
No credit card to try. Cancel anytime.
Free
Kick the tyres on a single deploy.
- 1 Quick Scan
- Core agent (Agent Smith)
- Functional + security-light
- Markdown export
Starter
For solo builders shipping weekly.
- 30 scans / month
- Standard + Deep Scan
- All 7 logical agents
- Fix prompt for Claude/Cursor
- Email support
Ship it. Then let Agent Smith test it.
Your next deploy deserves a QA pass. Run your first audit in under a minute.
Run your first audit